Phillips et al. have produced a fascinating paper, one that makes a strong case that knowledge representations should play a larger role in cognitive science than has occurred until now. As someone whose work in moral psychology has centered on moral knowledge, rather than moral beliefs, for over two decades (e.g., Mikhail, Reference Mikhail2000, Reference Mikhail2007, Reference Mikhail2011, Reference Mikhail2014), I am enthusiastic about their project. The questions they are asking, and the interdisciplinary methods they adopt, seem to me exactly the right approach to take to make progress in the theory of mind. I have a few quibbles, but because I am largely persuaded by their main argument, I wish here to accept it substantially as-is and to use this commentary to highlight additional lines of inquiry Phillips et al. might want to consider as they continue to develop this paradigm. All of them reflect the centrality of moral knowledge to moral psychology.

To set the stage, notice first how well moral knowledge fits many of the criteria outlined by the authors for determining whether some representations are more basic than others. For example, consider how natural it is to appeal to moral knowledge in ordinary conversation. We commonly refer to others as knowing the difference between right and wrong, rather than believing it. In a similar vein, we refer to others as knowing English, Hindi, Japanese, or other natural languages, rather than believing them. As a complex cognitive capacity, moral knowledge likewise typically emerges early in development (Hamlin, Reference Hamlin2013; Kagan & Lamb, Reference Kagan and Lamb1987), operates largely automatically in adults (Pizzaro & Bloom, Reference Pizzaro and Bloom2003), can be preserved in patients who suffer various other cognitive impairments (Nichols, Reference Nichols2004), and is shared to some extent with nonhuman primates (de Waal, Reference de Waal2006; Mikhail, Reference Mikhail2014). Unlike many beliefs, we do not forget our moral knowledge (Ryle, Reference Ryle and Meldon1958), and even a dog knows there is a moral difference between being stumbled over and being kicked (Holmes, Reference Holmes1991/1881).

Three further illustrations of the pivotal role played by moral knowledge in moral psychology seem worth highlighting in this context. Each of them suggests lines of inquiry that Phillips et al. or others may wish to pursue as they seek to deepen this promising research program. First, there is the explicit appeal to “conscience” as a datum of human nature, as manifested in the Universal Declaration of Human Rights and virtually all of the subsequent covenants and treaties that form the modern international law of human rights. What is conscience? The etymology of the original Latin is revealing here: con-scientia or “knowledge-with” – characteristically understood as knowing something with another (e.g., God, or one's inner self) (Potts, Reference Potts1980). Conscience is normally conceived to be a type of knowledge, not belief, and its attributions are historically and culturally ubiquitous. These facts and their behavioral effects may warrant further investigation within the diverse methodological frameworks proposed by Phillips et al.

Second, although they do not discuss this philosophical background, there are strong Kantian overtones to Phillips et al.'s claim that knowledge representations are more basic than belief representations. Indeed, the authors’ emphasis on knowledge representations appears at variance with the belief-desire psychology embraced by many contemporary philosophers and psychologists, the main elements of which often derive from a rival philosophical empiricism. For both moral cognition and other forms of cognition, the fundamental epistemological problem for Kant is: How is synthetic a priori knowledge possible? “Synthetic” and its counterpart, “analytic,” are adjectives that modify judgments or propositions, whereas “a priori” and its counterpart, “a posteriori,” are best understood as adverbs that modify verbs such as “to know” (Wolff, Reference Wolff1973). The key question, for Kant, is thus how synthetic judgments can be known a priori (i.e., prior to or independent of experience). There are many difficulties in interpreting Kant and applying his insights to modern cognitive science, of course, but one should not lose sight of the fact that a creative synthesis of Kant and Darwin is possible, in which the former's emphasis on core knowledge representations can be reinterpreted in evolutionary terms (e.g., Lorenz, Reference Lorenz and Ruse1941; Spelke, Lee, & Izard, Reference Spelke, Lee and Izard2010). The “call to arms” with which Phillips et al. conclude seems to me to lead most naturally in this direction, as does Phillips’ other interesting studies on causation, modality, moral judgment, and other topics (e.g., Phillips & Knobe, Reference Phillips and Knobe2018; Phillips, Morris, & Cushman, Reference Phillips, Morris and Cushman2019).

Finally, many familiar attributions of knowledge and ignorance in legal contexts also lend support to the principal argument advanced by Phillips et al. The clearest example may be the traditional maxim, ignorantia juris neminem excusat: “ignorance of the law is no excuse.” It reflects what is generally deemed to be an obvious fiction: that everyone is presumed to know the law. In an era in which thousands of obscure statutory or regulatory crimes can serve as the basis of criminal liability, this form of knowledge attribution might seem far-fetched and ridiculous. (For a frequently amusing and sometimes horrifying window into the full catalog of federal crimes, see the @CrimeADay Twitter feed.) A serious and substantive point lies behind the origin of this maxim, however, of which cognitive scientists should take note. Before the advent of modern statutory and regulatory crimes, everyone was presumed to know the law because the law generally reflected customary moral knowledge. Moreover, legally prohibited acts included, or were broadly similar to, those which researchers have recently discovered are condemned by human beings throughout the world, including non-WEIRD (western, educated, industrialized, rich, and democratic) populations in small-scale societies (Barrett et al., Reference Barrett, Bolyanatz, Crittenden, Fessler, Fitzpatrick, Gurven and Laurence2016; Fessler et al., Reference Fessler, Barrett, Kanovksy, Stich, Holbrook, Henrich and Laurence2015; Piazza & Sousa, Reference Piazza and Sousa2016; Saxe, Reference Saxe2016). These prohibitions, in other words, reflect core moral knowledge: a basic grasp of right and wrong, which can be validly denied only by those who fit the legal definition of insanity. Notably, in its most influential formula (the two-prong M'Naughten test), this definition is itself framed in terms of knowledge, rather than belief.

These observations merely scratch the surface of the many interesting possibilities opened up by Phillips et al. I look forward to seeing where their exciting cross-disciplinary research leads.